Backup destinations: NAS, Linux server or FTPS
Not every server can write to a CoreCP backup server. If you have a Synology NAS, a plain Linux server or an FTPS host, you can use it as a destination. The panel says plainly what that destination protects, and what it does not.
Written for: Administrator
Not every server can write to a CoreCP backup server. If you have a Synology NAS, a plain Linux server or an FTPS host, you can use it as a destination. The panel says plainly what that destination protects, and what it does not.
The band beside every destination
Wherever a destination is shown (the destinations list, a server's backup page, the fleet overview) you see the same band:
- Protected against deletion: a CoreCP backup server, or an SSH append-only destination where retention has been measured to run. A stolen key cannot erase anything here.
- Snapshots on the target (stated, not measured): you stated that the NAS or server keeps its own snapshots. The panel cannot check that, so it never calls it "protected".
- Not protected against deletion: everything else. Whoever can write can also delete. Enable snapshots on the target.
The kinds
| kind | for | protected |
|---|---|---|
| SFTP, profile Linux | a Linux server with an SFTP user | no |
| SFTP, profile Synology NAS | a Synology; the path is what the NAS shows (/backup/corecp), never /volume1/… | no, unless you state snapshots |
| SSH append-only | a Linux server you put a key on; the key may only append | yes, once retention runs there |
| FTPS | an FTPS host with a valid certificate | no; the lowest tier |
Plain FTP cannot be chosen: the password would travel in the clear. The panel says so.
Tried first, then saved
When you add a destination you pick a server that tries it: it connects, looks at the directory it lands in, writes a probe file and removes it, creates or opens restic's repository and writes and reads a probe snapshot. Every step is on a checklist. If one fails, nothing is saved and you see which step it was.
For SFTP and SSH the panel shows the exact line to paste into authorized_keys on the target. Copy it literally: for SSH append-only the protection is in that line.
SSH append-only: retention
On such a destination the server may only append, so old backups are removed on the destination itself. When you link the server, the panel asks you for a prune-key password: choose one and keep it yourself. The panel passes it once to the server and does not keep it. On the target you run the setup script the panel hands out once; it asks for the same password and installs a daily timer.
After every backup the server checks that retention still runs there. If the timer has stopped you see retention is not running on the destination beside the server and as an attention item, and the band reads not protected until it runs again.
FTPS: the weekly check
FTP has no checksums. So on an FTPS destination the server reads part of the data back every week after retention. A damaged file is a red row beside the server, not a warning, and an attention item.
The exception per server group
A destination that is not protected against deletion can only be attached to a server in a server group for which a server administrator has set the exception, with step-up and a reason. Without it the panel refuses the attach and says why.
Groups that already used such a destination when this rule arrived got the exception as "pre-existing" plus one attention item: confirm it with a reason, or move the servers to a protected destination. Nothing that was backing up has stopped. If you withdraw the exception later, new attachments are refused; what is attached keeps running until you detach it yourself.
From the terminal
corectl backup ssh-key --target ssh-append-only --repo /srv/restic/w1 # the key line and the retention script
corectl backup destination check --target ftps --endpoint ftps://nas.example.net --repo backups/w1 --ftps-user backup --ftps-password-stdin
corectl backup check # the integrity read, now
corectl backup status --json | python3 -c 'import json,sys; print(json.load(sys.stdin).get("target_retention"))'See also
- Your servers' backups
- Restoring a backup